Using the instructional computer simulation "Hunger in the Sahel", two experiments were conducted concerning the moderating effect of domain knowledge on the correlation of intelligence and problem solving. Experiment 1 with N = 200 students implemented a between-subjects design, Experiment 2 with N = 28 young adults a within-subjects design with 10 repeated measures on problem solving. The results correspond to the Elshout-Raaheim hypothesis: With low domain knowledge, the correlation is low; with increasing knowledge, the correlation increases; with further increasing knowledge, the correlation decreases; finally, when the problem has become a simple task, the correlation is again low. The results are of practical and theoretical relevance for designing simulation-based learning environments and simulation-based tests for measuring intelligence and problem-solving ability.